Introduction - If you have any usage issues, please Google them yourself
The basic process k center algorithm is: First free to choose a delegate object for each cluster, the rest of the object based on its distance each of which represents the object (here the distance is not necessarily a Euclidean distance, Manhattan distance may be) allocated to Recently representatives object represents a cluster and then repeatedly with representatives of non-objects instead represents the object, in order to optimize the quality of clustering. Clustering quality is represented by a cost function. When a center is a non-center alternative, in addition to not replace the center, the rest of the points to be reallocated.